A primal neural network for solving nonlinear equations and inequalities

摘要

In this paper, the concept and utility of primal neural network are introduced for the context of dynamical constraints or inequalities. Based on the neural-network design experience on solving linear equationdinequalities, we generalize a primal neural network to handling the nonlinear situation. Numerical examples (including the robotic applications) are given to demonstrate the effectiveness of the primal network.